Meaning After Work

There is a particular anxiety that the current wave of AI produces, distinct from the fear of losing a paycheck. It is the fear of being made pointless. For a couple of centuries, machines took over our muscles, and we consoled ourselves that the human contribution had simply moved upstairs, into the mind — into judgement, creativity, care, the things a lever could never do. Now the machines are climbing the stairs. The unspoken question underneath the headlines is not only will I have a job? It is if a machine can write the essay, compose the tune, draft the diagnosis — what am I for?

A confusion worth naming

The question feels devastating because it smuggles in an assumption: that a human being's worth is a function of what they can produce that nothing else can. Call it the productivity theory of meaning. On this view, you matter to the exact extent that you are useful, and useful means: capable of outputs the market will pay for. It is a theory almost nobody would endorse out loud — no one believes a child, a retiree, or a friend earns their standing by output — and yet it is the theory doing the frightening. The dread is real; the premise underneath it is quietly false.

We have confused the tasks we do with the point of doing them. Automate the first and you have not touched the second — unless you had already let the first stand in for it.

Keynes’s missing hour

In 1930, in the depth of the Depression, John Maynard Keynes wrote a curious little essay, Economic Possibilities for Our Grandchildren. He predicted, roughly correctly, that within a century productivity would rise so far that the economic problem — mere subsistence — would be essentially solved, and his grandchildren might work fifteen hours a week. He was not naive about it. He saw the catch clearly. The problem, he wrote, would no longer be how to produce enough, but how to live: how a species trained too long to strive and not to enjoy would fill the hours once survival stopped demanding them.

The productivity part of his prediction mostly came true; the leisure part did not, because we found new wants faster than we automated old work. But the deep point survives untouched, and AI sharpens it. Keynes saw that work had been doing two jobs at once. It produced the goods we need, and it structured our days, our status, our sense of contributing. Solve the first job and the second is suddenly exposed, unsupported, asking to be answered directly instead of as a side effect of the paycheck.

Where meaning actually comes from

Strip away productivity and the sources of meaning that remain are not new; they are the ones philosophers kept pointing to before industrial work ever colonised our self-image. They share a feature worth noticing: none of them is threatened by a machine doing the same task faster.

• The doing, not the done. Much of what makes an activity meaningful lives in the process — the absorption of playing music, the friction of learning a language, the satisfaction of a problem worked through by hand. That a machine can generate the output in a second removes none of the good of the doing. A chess engine crushed the best humans thirty years ago, and more people play chess now than ever. We did not play to produce moves the world lacked. We played to play.
• Relationship. To be known by particular people, and to know them; to be depended on; to give and receive care. This is arguably the largest single source of a life feeling worthwhile, and a machine that can draft your emails does nothing to it, for good or ill. It is not a productivity problem, so productivity tools cannot solve it or spoil it.
• Contribution to something you did not have to. Meaning tends to grow where our effort meets a genuine need — a person helped, a thing repaired, a small corner of the world left better. Automation changes which efforts are needed. It does not abolish the standing human fact that being of use to others, freely, is one of the most reliable ways to feel your life has weight.

The pattern is plain once you look. Meaning is not primarily about being irreplaceable. It is about being engaged — present to your own life, bound to other people, spending yourself on things you judge to be worth it. A being can be economically redundant and richly, unmistakably alive. We already know this, because we already believe it about everyone we love who is not currently at work.

What happened when the work went away

Everything above is argument. There is also ninety years of evidence on what happens when paid work leaves people’s lives, and read one way it refutes this essay. It deserves a hearing first.

Marienthal, 1931–32. The textile mill at Marienthal, a workers’ village near Vienna, had shut by 1930. From November 1931 to May 1932 a team of 17 led by Paul Lazarsfeld, Marie Jahoda and Hans Zeisel studied the village, publishing Die Arbeitslosen von Marienthal (The Unemployed of Marienthal) in 1933. What follows is from accounts of the book, not the book itself: the historian Joseph Malherek’s summary of the study and the House of Austrian History’s entry on it. The researchers kept records on 478 families and time sheets for eighty people. The week stopped turning on Sunday and started turning on the fortnightly relief payment. Women stayed busy running households and children kept the school week, while the unemployed men, in Malherek’s words, simply drifted along in a timeless, undisciplined existence. They read less, not more. Subscriptions to the workers’ newspaper fell by more than sixty percent from 1927, and about seventy percent of families were classed as resigned. Where the left expected revolt, the researchers found what the House of Austrian History calls a tired community marked by resignation.

What a job gives besides pay. Jahoda spent decades asking why. Her answer, in Employment and Unemployment (1982) and summarised in a chapter by Brendan Burchell and Alex J. Wood, is that income is only the manifest function of employment. A job also has five latent functions, kinds of experience that meet psychological needs: time structure, enforced activity, social contact outside the family, collective purpose, and status and identity. This is the Keynes point above, that work was doing two jobs at once, with the second job itemised. That makes it a checklist: when a job goes, each of the five has to come from somewhere else, or not at all.

The size of the harm. Karsten Paul and Klaus Moser’s meta-analysis, “Unemployment impairs mental health” (Journal of Vocational Behavior, 2009), pooled 237 cross-sectional and 87 longitudinal studies. On average 34% of unemployed people had psychological problems, against 16% of employed people. According to its abstract, longitudinal studies and natural experiments endorsed the assumption that unemployment is not only correlated to distress but also causes it. Men and blue-collar workers suffered more, and the harm was larger in poorer countries, in countries with unequal incomes, and in those with weak unemployment protection systems.

Finland, 2017–18. For two years Finland’s social insurance institution, Kela, paid 2,000 randomly selected unemployed people a tax-exempt €560 a month, regardless of any other income they may have had or whether they were actively looking for work. A control group of other unemployed people received the ordinary benefit. The final results came in a joint press release from Kela and the Ministry of Social Affairs and Health (6 May 2020). The effect on employment was small: six extra days of work in the measured year, and no effect at group level in the first year. The effect on wellbeing was clearer: Survey respondents who received a basic income described their wellbeing more positively than respondents in the control group. They were more satisfied with their lives and experienced less mental strain, depression, sadness and loneliness. Minna Ylikännö, head of the research team at Kela, added the caveat in the same release: On the basis of the study, it is not possible to say with certainty that the better wellbeing of the test group was due specifically to receipt of a basic income. Only 23% of those surveyed responded.

Be clear about what the trial was. Its participants were unemployed people given modest unconditional money for two years, not people freed from needed work among neighbours who had also stopped. It tested whether dropping conditions from benefits helps the unemployed. It cannot test a world without work.

The case against this essay, at full strength. Take the job away, this reading says, and people do not blossom into musicians and friends. Marienthal’s men had every hour free and a workers’ library, and they drifted and read less. The meta-analysis says unemployment causes distress, not merely accompanies it. Jahoda herself held, as Burchell and Wood report, that unpaid work cannot adequately provide status and identity, or a share in society’s collective purposes, without employment. She also wrote that most people cannot single-handedly overthrow the compelling social norms under which we all live, adding: There are at all times only a few who can manage without it. Lazarsfeld later suggested, as Malherek reports, that the study’s most important conclusion was perhaps that the dole is wrong, and work relief the better answer. On this reading, the “sources of meaning that remain” are a philosopher’s list, and a job is where most people actually find them.

What the data can and cannot separate. Every study above removes several things at once: income, status, routine, company and purpose leave together, and shame often arrives. Where the evidence can pull them apart, money and security explain a large share of the harm. The Marienthal authors themselves traced a link between a family’s income and its attitude to life, as Antoni Sułek notes. Jahoda later gave the size and permanence of benefits as the main reason why the unemployed she studied in Monmouthshire, Wales, in 1937–38 suffered less than Marienthal’s. Paul and Moser found the harm worse where protection is weak. Finland’s recipients reported better wellbeing and a more positive view of their economic welfare, with almost no change in employment. But the rest matters too: in the studies Burchell and Wood review, social contact and time structure are among the functions most often tied to wellbeing, alongside financial strain. What no study can test is the case this essay is about: a society where not having a job brings no loss of income or standing, and most others have stopped too. None has existed.

So the verdict is split. The evidence supports this essay’s claim that much of the damage is about how income and standing are distributed, and refutes any hope that the rest will take care of itself. A job supplied structure, activity, company, purpose and standing without anyone building them. Without it they must be built on purpose, by people and institutions, or you get Marienthal.

The real risk, stated honestly

This is not a counsel of complacency. There is a real danger here, but it is not the metaphysical one. The danger is not that machines make human life pointless. It is that we have arranged a society in which access to income, status, and self-respect flows through paid work, so that people thrown out of the labour market lose those goods even though their capacity for a meaningful life is fully intact. That is a failure of distribution and imagination, not of human worth. The evidence in the previous section backs the distribution half of that claim and adds a warning to the imagination half. Income alone will not rebuild the time structure, company and shared purpose that a job used to supply as a side effect; those have to be designed. It is solvable, and it is on us to solve, and it will not solve itself by anyone insisting the old arrangement was natural.

Keynes thought the transition would be hard precisely because we would have to unlearn habits of striving that had been, for millennia, simple necessities. He was right about the difficulty and, so far, early on the timing. But he pointed at the correct task, and AI has only made it more pressing: to decouple the question am I productive? from the question is my life worth living? — and to build, deliberately, the kind of society in which the honest answer to the second can be yes regardless of the first.

The machines are not coming for our meaning. They are calling a very old bluff — the one where we let our jobs answer, on our behalf, a question we were always going to have to answer ourselves.